Senior LLM Engineer (AI)

Location: Barcelona, Spain; hybrid workplace

Company

hirify.global is a well-funded deep-tech company founded in 2019 that develops quantum software and AI products, including CompactifAI for foundational-model compression and Singularity for quantum and quantum-inspired optimization.

What you will do

• Train and post-train LLMs and multimodal models on multi-node GPU clusters using SFT, preference optimization, RL, and distillation.
• Design model compression, quantization, pruning, quantization-aware training, tensor-network methods, and healing recipes.
• Build RL environments, reward functions, and verifiers for reasoning, coding, agentic tasks, world models, and Physical AI.
• Research post-transformer and hybrid architectures, including state-space models, linear attention, and MoE.
• Build large-scale data pipelines, synthetic-data generation workflows, benchmark evaluations, and internal evaluations.
• Optimize models for inference on accelerators and edge chips, document results, review code, mentor teammates, and contribute to papers or open-source releases.

Requirements

• Master’s or PhD in Computer Science, AI, Mathematics, Physics, or a related field, or equivalent industry experience.
• 4+ years of hands-on experience training or post-training transformer models.
• Strong PyTorch skills, including model code, custom losses, and parallelism configuration below the library level.
• Experience with distributed training using FSDP, DeepSpeed, or Megatron-LM; data, tensor, pipeline, and expert parallelism; and multi-node clusters with Slurm or Kubernetes.
• Hands-on experience with SFT, DPO, and at least one RL method such as PPO or GRPO.
• Strong foundations in linear algebra, optimization, probability, and training dynamics.
• Good understanding of GPU memory hierarchy, mixed precision, profiling, and throughput.
• Rigorous experimental practice covering ablation design, reproducibility, experiment tracking, and noise-aware evaluation.
• Strong debugging and communication skills; fluent English required.

Nice to have

• Experience with NVIDIA frameworks or similar RL and training frameworks, including Megatron Bridge, NeMo RL, NeMo Gym, NeMo Curator, verl, TRL, or OpenRLHF.
• RL infrastructure experience with rollout engines, asynchronous RL, reward models, verifiers, and agentic environments.
• Experience with post-training quantization, QAT, pruning, distillation, low-rank methods, or tensor-network methods.
• Experience with post-transformer architectures, Physical AI, robotics simulation, or multimodal model training.
• Inference experience with alternative hardware and tools such as ExecuTorch, llama.cpp, ONNX Runtime, OpenVINO, vendor NPU SDKs, Triton, or CUDA.
• Experience with large-scale data processing, synthetic-data generation, evaluation tooling, or Artificial Analysis.
• Publications at major research venues, notable open-source contributions, or models released on Hugging Face.
• Spanish language skills.

Culture & Benefits

• Work in a multicultural and international team of more than 180 employees.
• Join a fast-growing, well-funded deep-tech company focused on quantum computing and artificial intelligence.
• Research priorities change quickly, with an emphasis on strong fundamentals over narrow specialization.
• Opportunities to publish research papers and release models and code when appropriate.